Sie sind auf Seite 1von 16

Scaling of Health Information Systems in India:

Challenges and Approaches


Sundeep Sahay
Information Systems Group, Department of Informatics, University of Oslo,
Gaustadalleen 23, P.O. Box 1080, Blindern NO-0316, Oslo, Norway.
E-mail: sundeeps@ifi.uio.no
Geoff Walsham
Judge Business School, University of Cambridge, Trumpington Street,
Cambridge CB2 1AG, United Kingdom.
E-mail: g.walsham@jbs.cam.ac.uk

ABSTRACT

This article addresses the issue of scaling of information systems (IS) in both theoretical and empirical
terms. Scaling is an important issue in IS, especially in the contemporary context of globalization,
as attempts are ongoing to expand IS in the same context as well as take it into other contexts.
Theoretically, an information-infrastructure (II) perspective is drawn on to analyze the challenge
of scaling, viewing it not merely as a technical problem, but as a socio-technical one involving a
heterogeneous network constituted of technology, people, processes, and the institutional context.
Empirically, scaling is analyzed based on experiences from an ongoing project to implement health
information systems within the primary health care sector in India. The theoretically informed
empirical analysis leads to some preliminary insights relating to the questions of what is being scaled
and how it is being scaled. Some conclusions are drawn on theoretical and practical challenges related
to scaling, and on implications for human-resource capacity development.  C 2006 Wiley Periodicals,
Inc.
Keywords: scaling; health information systems; India; globalization; information infrastructure

1. INTRODUCTION
Although in practical terms scale refers to the size or scope of something (for example, an
information system or a process), scaling concerns the process through which that product
or process is taken from one setting and expanded in size and scope within that same setting
and/or also incorporated within other settings. In the context of information systems (IS),
scale then could refer to the scope of an IS (for example, how many users are served),
whereas scaling could imply the expansion of this system in scope and size (for example,
making the system accessible to more users or increasing its functionalities). Although a
number of contemporary debates in IS research point to the problem of such expansion,

Abiodun O. Bada was the accepting Special Issue Editor for this paper.

Information Technology for Development, Vol. 12 (3) 185–200 (2006) 


C 2006 Wiley Periodicals, Inc.
Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/itdj.20041

185
186 SAHAY AND WALSHAM

the notions of scale and scaling are hardly ever explicitly discussed in IS research. When
issues of scale are discussed (for example, Monteiro, 1998), the focus tends to be on the
technical artifacts, and rarely are social issues, for example, the human resources capacity
to deal with the enhanced scope of the systems, explicitly considered.
The issues of scale and scaling have particular implications in the context of health care
and health information systems (HIS) in developing countries, which is the empirical focus
of this article. For example, various reform efforts related to the introduction of computer-
based HIS have died premature deaths as “pilot projects” (Heeks, Mundy, & Salazar, 1999)
because they could not be expanded to a level where they were useful for health managers
(Braa, Monteiro, & Sahay, 2004). For example, to make effective decisions on resource
allocation for a district, the manager needs health data from all the clinics in the district and
not just from an isolated and limited set of clinics from a preselected pilot area (Braa et al.,
2004). However, for these systems to grow to this required level of scale, they need to be
accompanied by the scaling of human resources capacity, at two levels at least. The first is
at the level of users. The growth in scale of a HIS is often is associated with an escalation
of its technical complexity, for example, through the need for larger databases, greater
bandwidth, and the increased need to interface with existing legacy systems (Kimaro &
Nhampossa, 2005). This increased complexity requires that users have a higher level of
technical competence in order to be able to use these systems effectively. Secondly, human
resource capacity needs to be scaled at the level of the implementation team responsible
for providing technical and organizational support to the users and the user organization.
With increased scale, the implementation team needs to address a wider geographical area,
deal with more complex technical problems, and also be increasingly involved with issues
that are not just technical in nature, but also institutional and political. This requires the
implementation team capacity to be scaled up both in terms of numbers and the skills they
possess. This article argues that the scaling challenge is inextricably linked with issues
of human resources capacity, a crucial consideration in the implementation of any health
reform effort in the context of developing countries (Kohlemainen-Aitken, 2004).
More broadly than the health sector, the notion of scaling is also relevant to contempo-
rary debates around globalization, but again this link is rarely discussed explicitly by IS
researchers. For example, a homogenization perspective argues that globalization is leading
to a global convergence of processes and products, implicitly assuming that similar systems
can be scaled up; that is, they can be expanded in scope and incorporated in other settings.
The opposing perspective of heterogeneity (for example, Appadurai, 1996), which argues
that different groups will appropriate globalization processes based on their socio-political-
historical contexts, sees scaling of systems to be inherently problematic. A middle ground
on these debates represented by the “glocalization” perspective (Robertson, 1992) views
the local and global as constituted of and constituting each other. Robertson points out
that glocalization finds its origin in the Japanese word dochakuka, roughly meaning global
localization, because of the Japanese concern for incorporating local consensus which then
needs to be scaled up within a global framework. This middle-ground view is also reflected
in the Rolland and Monteiro (2002) argument for a “pragmatic balance,” where no attempt
is made to scale up systems unproblematically under the assumption of “one size fits all.”
Instead careful examination is made of which aspects of the systems can be scaled up and
which require local customization.
The present argument is that achieving this pragmatic balance explicitly requires dealing
with the questions of scale (the level of the balance) and scaling (the process through which
this balance is achieved and maintained). This requires an approach that considers, in an

Information Technology for Development DOI: 10.1002/itdj


SCALING OF HEALTH INFORMATION SYSTEMS IN INDIA 187

interconnected manner, aspects of technology, people, processes, and the institutional


context within which they are embedded. The article thus draws upon the sociology of
technology literature (Callon & Law, 1986; Latour, 1987) so as to treat scale not as only
a technical issue (such as large-scale systems and architecture) or an economic matter
(economies of scale or of increasing returns), but as a socio-technical issue. More specif-
ically, the article draws upon the information infrastructure perspective (Hanseth & Mon-
teiro, 2004; Star & Ruhleder, 1996), which is specifically concerned with large (scale),
interconnected, complex systems.
This article has two research aims: one, to develop a deeper conceptual understanding
of the problem of scale; and two, to apply this understanding to an empirical setting.
The empirical basis for the analysis concerns a complex and interconnected system for
primary health care (PHC) in India. Problems of scale and scaling are at the heart of
the implementation challenge of health information systems in PHCs (Braa et al., 2004),
especially in India. The rest of this article is organized as follows. Section 2 develops the
theoretical perspective, and the case study is presented in Section 3. Section 4 includes
the case analysis, drawing upon the theoretical perspective presented in Section 2. Finally,
some implications of the analysis are presented in Section 5.

2. SCALING AND INFORMATION SYSTEMS:


SOME THEORETICAL CONSIDERATIONS
The problem of scaling has not received the attention required by IS researchers. Monteiro
(1998) provides a quote from the IITA (1995) report to emphasize this point:
We don’t know how to approach scaling as a research problem, other than to build
upon experience with the Internet. However, attention to scaling as a research theme is
essential and may help in further clarifying infrastructure needs and priorities. . . It is clear
that limited deployment of prototype systems will not work. . .
Recent research in IS around information infrastructures (II) provides a promising per-
spective to theoretically unpack the issue of scale. Information infrastructures, for example,
the Internet, represent integrated solutions based on an ongoing fusion of ICTs and pro-
cesses. Rolland (2003) defines an II as follows:
“A heterogeneous collection of different technologies, components, protocols and appli-
cations to support different and varying applications areas and use over time across large
geographical distances.”
Hanseth and Monteiro (2004) describe IIs to be characterized by three features:

Shared: II is one irreducible unit that is shared by a larger community of users, as a


part of the II cannot be reduced and used independently without calling into play the
other interconnected parts.
Open: IIs are open in the sense that there are no predefined limits to the number of
users and stakeholders. Without necessarily implying that everyone and anyone is
included in an II, they argue that predefined boundaries to IIs cannot be created.
Enabling: IIs, rather than being designed for particular singular purposes, should be
seen as playing a supporting or enabling function for a variety of different application
domains.

The II perspective helps to emphasize that the social and technical are not separable and
are instead constituted and constitutive of one another. For example, Latour (1999) argues
Information Technology for Development DOI: 10.1002/itdj
188 SAHAY AND WALSHAM

that “airplanes do not fly, airlines do,” implying that the artifact of the plane does not fly
on its own, but requires a complex and heterogeneous socio-technical network comprising
of pilots, navigators, runway staff, air-traffic–control towers, radars, runways, and flight
schedules. Viewed from this perspective, scale is not just a technical or economic issue,
but one of scaling a heterogeneous and complex network. However, the goal herein is to go
beyond the Hanseth and Monteiro focus on what an II is to also examine the socio-technical
processes and embedded practices by which the II is constructed. The issue of scaling is
fundamental to this process.
The cultivation approach to II design described by various proponents of the II per-
spective (for example, Ellingsen & Monteiro, 2003; Hanseth & Aanestad, 2003; Hanseth
& Monteiro, 2004; Rolland, 2003) provides a rich analytical tool with which to explore
the socio-technical processes and embedded practices that shape the scaling problem. The
cultivation approach represents a more conservative approach to design than construction,
which tends to emphasize the power of human agency in “selecting, putting together, and
arranging a number of objects to form a system” (Dahlbom & Janlert, 1996, p. 6). Instead,
cultivation emphasizes the power of the material; “the tomatoes themselves must grow,
just as the wound itself must heal. . . ” (Dahlbom & Janlert, 1996, p. 6), implying that
the “development organization” or “product” being developed should be considered as a
unified socio-technical network; neither should be prioritized over the other.
The power of the material, which the cultivation approach emphasizes, relates to the II
concepts of the installed base and the resulting lock-in effects (Hanseth & Monteiro, 2004).
A classical example in this regard is the QWERTY keyboard, which represents an installed
base on which the current design of computer keyboards is still based. This lock-in effect
represents a dilemma in the scaling of an II. Using the example of the Internet, Monteiro
(1998) describes how the expansion of the Internet creates new patterns of use, whereas

TABLE 1. Summary of Theoretical Perspective

Concept Description
Scaling Scaling is not only about numbers and size (although these are
important elements of the problem), which can be achieved through
network externalities (where as the value of a technology increases,
more users will adopt it), but refers to the processes and embedded
practices by which heterogeneous networks around the technology
are spread, enhanced, scoped, and enlarged. Thus scaling concerns
aspects of geography, software architecture, people, processes,
infrastructure, technical support, and political support.
Cultivation & approach A cultivation approach acknowledges the existence of the installed
base and the lock-in effects and represents an appropriate approach
to tackle the challenge of scaling up a complex interconnected
system. This approach seeks to address change in an incremental
and gradual manner, changing small parts while maintaining
alignment with the rest of the network.
Unanticipated & effects In any change effort, there are both anticipated and unanticipated
effects. Anticipated effects are best understood in localized
conditions, but how these may propagate to the larger network is
difficult to predetermine. Unanticipated effects, arising from the
interconnected nature of the information infrastructure and
incomplete knowledge of the whole, are inherent in the problem of
scaling, and can contribute to both challenges and opportunities.

Information Technology for Development DOI: 10.1002/itdj


SCALING OF HEALTH INFORMATION SYSTEMS IN INDIA 189

the infrastructure itself has a strong, conservative influence (arising from a large installed
base of routers, users’ experience and practices, backbones, hosts, and specifications) that
favors a situation of inertia and challenges the scaling-up processes. This dilemma cautions
against the need to adopt radical (construction) approaches to change and instead favors
a smooth and incremental (cultivation) strategy that involves changing one small part at a
time while keeping the changes simultaneously aligned with the rest of the network.
A cultivation approach to scaling emphasizes the “improvisational” processes of change,
and the potential of what people do in situated action (Suchman, 1987), and does not just
focus on planned and rational approaches (Ciborra et al., 2000). Design is seen not as a
well-defined process with preconfigured start and end states, but as an ongoing process of
ecological change. The interconnected nature of the II creates the potential for “unantici-
pated effects” (Walsham, 1993) or “drift” (Ciborra et al., 2000), which reflect the inability
to anticipate events in advance. This helps to emphasize the need to adopt an approach that
is in small steps, incremental, and considers flexibility and change. Hanseth and Aanestad
(2003) use the bootstrapping metaphor to describe such a design strategy in the analysis of
a telemedicine application conceptualized as an II.
A summary of the theoretical perspective is given in Table 1.

3. CASE STUDY OF HEALTH INFORMATION SYSTEMS IN INDIA


The empirical material derives from a project relating to the design, development, and
implementation of the Health Information Systems Project (HISP) (Braa & Hedberg, 2002)
in Andhra Pradesh (AP), a state in Southern India. AP, by its sheer size and population
levels, represents a complex problem of scale, including a population of about 75 million
spread over an area of 246,793 sq km divided into 23 districts for administrative pur-
poses. A network of 1,500 primary health centers (PHCs) and 7,500 subcenters (SCs) are
responsible for catering to the health-care needs of the population. The case study is de-
scribed in three phases, reflecting three different and increasing levels of scale: the Kuppam
pilot, the Madnapally expansion, and the statewide expansion. Table 2 provides a summary
of the different project phases, and these phases are described in more detail in the text,
with a focus on issues of scaling.

3.1 The Kuppam Pilot Phase


After permission was received from the state and district levels in 2000, the field work was
begun in the Chittoor District, which has a population of about 3.75 million and 84 PHCs.
Initially, a situation analysis was conducted in order to obtain an initial understanding of
the organizational structure and patterns of health information flows. Then a minimum data
set (MDS) of items that can be used by the PHCs and SCs was designed. This analysis
was done through field work conducted over a year in 2001 when the researchers visited
a number of the health facilities and had discussions with various health staff including
field-level workers, administrative staff, and medical doctors. In-depth study of various
report formats, copies of registers, and diaries used for data collection was also performed.
Two outputs resulted from this study. The first concerned the design of a MDS which
sought to rationalize the data elements being collected by the PHCs and SCs by identifying
the minimum set of common items. The second output concerned the conceptualization of
the structure of the health information flows as depicted in Figure 1. As Figure 1 depicts,
Information Technology for Development DOI: 10.1002/itdj
190 SAHAY AND WALSHAM

TABLE 2. Summary of Project Phases

Phase Period Key activities


The Kuppam December 2000– Situation analysis at the district office to
pilot phase September 2002 understand information flows.
Definition of minimum data sets.
Implementation through HISP of the District
Health Information System (DHIS) in 12 PHCs
of Kuppam constituency.
Intensive and continuous training of health staff.
The Madnapally January 2003–July 2003 Expanding the Kuppam systems to Madnapally
expansion phase revenue division, consisting of 46 PHCs.
The use of a cluster-based strategy to address
some of the limitations experienced in Kuppam.
Enhancement of the technical systems, for
example, the incorporation of Geographic
Information Systems and Web-based systems.
State-level January 2004–May 2004 Establishing the DHIS in the DMHO offices of all
expansion 23 district offices.
Setting up a state-level database on MySQL
platform to import monthly data from all
districts.
Web-enabling some of the systems to enhance
visibility of the information.
Significantly increasing the size of the HISP
implementation team.

Figure 1 Overview of health information flows.

Information Technology for Development DOI: 10.1002/itdj


SCALING OF HEALTH INFORMATION SYSTEMS IN INDIA 191

the subcenter, the facility closest to the community providing health care, has as its parent
a particular PHC to which it sends its monthly reports. The PHC then in turn aggregates
all the reports received from all the subcenters under it, and sends this to the District
Medical Health Officer’s office, which in turn collates the PHC reports and sends them
to the Commissioner of Family Welfare’s office at the state level. In addition to these
routine health information flows, health programs like Malaria and Leprosy have their own
information systems and report directly to the state health program offices, often bypassing
the DMHO office. Similarly, the hospital system has its own parallel reporting system that
is independent of the DMHO.
As Figure 1 illustrates, there are a number of structural issues that shape the information
flows. First, the fragmented and vertical data flow results in data redundancies and poor
integration of information at the district level that is supposed to serve as the hub of
this information network. Second, there is a steady aggregation that takes place as the
data moves up the hierarchy, which systematically masks the facility-level data, making
it difficult to analyze the situation at the local level. Third, the arrows are primarily one-
way (bottom to top), reflecting limited feedback of information to support the local levels.
Fourth, the separation between the hospital and PHC sector represents an obstacle to unified
management of health services at the district and subdistrict levels.
In September 2001, the results of the analysis were presented to the Commissioner of
Family Welfare (CFW), who is the overall head of the health department in the state,
and also to the Chief Minister (CM), who is the highest political figure in the state. The
CFW, who was interested in supporting an alternative World Bank–funded project, was not
very amenable to these efforts. The CFW’s resistance was addressed by leveraging on the
support of the CM’s office. A presentation to the CM was made in the presence of other
senior officials from the health and IT departments. The CM was very appreciative of these
efforts, and sanctioned 12 computers for implementing the program in the nine PHCs in
Kuppam constituency1 (plus three for the district office). Out of the nine PHCs, three of the
computers had to be subsequently withdrawn because of infrastructure problems related
to the lack of a proper building, lack of staff, or security concerns. The implementation
was thus conducted in 6 PHCs. The initial months of the project were severely impeded
by poor infrastructure and poor technical support. The computers or the power supply
constantly blew up because of power fluctuations, caused by the absence of earthing wires.
This problem was further magnified by poor technical support, because the vendors were
reluctant to travel to the distant rural areas, especially after having received their payment
for the computer supply. Disruption in power supplies, sometimes up to 10–12 hours a day,
further impeded the progress of the project.
To deal with the local challenges, a partnership with a local computer company was
developed to provide training and local support to the PHCs. In the first 3 months, this
company provided full-time support in the six PHCs in addition to conducting a monthly
combined workshop for staff from all the nine PHCs in a central location. After 3 months,
one trainer was made responsible for supporting all the nine PHCs for an additional 9
months. The process of training was accompanied by the task of customizing the software
to local needs, including the implementation of the MDS, populating the database, and
automation of the routine reports required to be sent monthly from the PHC to the state. The

1An electoral constituency represents a geographical area for electing a member to the state assembly. Kuppam

is one such constituency within Chittoor district from which the CM had been elected in the past. This geographical
area consisted of nine PHCs where the initiative was implemented.

Information Technology for Development DOI: 10.1002/itdj


192 SAHAY AND WALSHAM

ongoing feedback received from the health staff and also through the combined workshops,
helped to continually improve the systems, especially the data sets and report formats.
In September 2002, an official from the CFW office along with some district officials
visited Kuppam to evaluate the project. They visited some of the PHCs and saw how
data-entry work was being carried out by the health staff, who were also asked to explain
various features in the software relating to exporting data, making graphs and charts, and
generating reports. Ten days after the evaluation, it was learned that a Government Order had
been sanctioned to extend the project to the Madnapally revenue division, which included
46 PHCs, and a memoradum of understanding (MoU) was subsequently signed between
the CFW and HISP. Through this MoU, the Government sanctioned US$35,000 for the
purchase of computers, trainer salaries, and to support development costs.

3.2 The Madnapally Phase: Expanding to 46 PHCs


The MoU mandated the expansion of the DHIS and related HISP processes to the 46 PHCs
of Madnapally revenue division (which also included the 9 PHCs of Kuppam) within the
relatively short time of 6 months. This presented a significant challenge of scale at least
at two levels. First, the HISP team had to be scaled up from the two full-time members to
nine, which posed challenges of getting the staff up to speed both with the HISP software
and the workings of the PHC. Second, it was necessary to expand the size and scale of the
operations, including the number of PHCs covered, to train a larger health staff, and to cope
with the larger distances between the PHCs.
Lessons learned from the Kuppam experience led the researchers to address the scale
challenge first in structural terms. Because many of the PHCs did not have adequate
capacity and infrastructure to host the computer, a cluster strategy was adopted, whereby a
computer would not be placed in every PHC, but in a hub PHC that would serve a cluster
of four–five geographically proximate PHCs. The hubs were located in what is called as
a 24-hour PHC, which typically had relatively stable infrastructure (for example, power
supply), and a secure room in which the computer could be placed. The HISP trainer was
based in this hub, and he or she established schedules whereby the health staff from the
cluster would come to receive training, to do data entry, and to generate monthly reports.
To address the challenges of power fluctuations and the damages caused by it, an audit of
the electrical wiring was conducted before the computer was installed, and the earthing
was replaced where required. Dot-matrix printers replaced laser-jets ones, as the dot-matrix
printers require less support and fewer consumables. A hardware engineer was among the
eight new recruits, to strengthen the provision of hardware support and to help develop
local competence.
This cluster strategy generally worked, although some of staff from the nonhub PHCs
protested against not having computers themselves. Within 6 months levels of implemen-
tation (training, database population, report generation) were similar to those achieved in
Kuppam; although in Kuppam this the level of implementation had taken about double the
time to achieve. Many of the PHCs started to generate their own monthly reports with the
software, and started presenting them to the district office instead of the manually generated
reports presented in the past. In the meantime, the project started to face tensions arising
from the State’s efforts to implement another software system called FHIMS2 (Family

2
FHIMS is a software system that enables the collection of data based on individual names of patients, as
contrasted with the HISP software, the DHIS, that deals with aggregated data for facility-based (PHC and SC)
reports.
Information Technology for Development DOI: 10.1002/itdj
SCALING OF HEALTH INFORMATION SYSTEMS IN INDIA 193

Health Information Monitoring System). As a part of this project, in September 2003, com-
puters were installed in all the PHCs in the state (about 1,500), and the FHIMS software
was expected to be installed within the next 6 months. The health workers started to get
conflicting signals about whether they should continue with the HISP approach or wait
for FHIMS. In some cases, where there was a good personal rapport between the HISP
trainers and the PHC staff, the HISP processes continued; in others the momentum started
to be lost. Also, because of the danger of being thrown out due to FHIMS, the need for a
strategy of integration rather than competition was determined, and a proposal for this was
made to the CFW. The CFW was not positive about this proposal, preferring unambiguous
statewide FHIMS implementation. However, with much persuasion and support from the
Chief Minister’s office, she agreed to a contract to implement a new Web-based system for
the monitoring of infant and maternal mortality deaths in all the 23 districts in the state.
This project was seen as being relatively noninterfering with the FHIMS implementation,
and was also to take place at the district rather than the PHC level (where FHIMS was
currently focused). However, along with this project, an agreement was reached with the
CFW for the development of a district-level database for all 23 districts, and for linking the
routine data to maps with the use of geographical information systems.

3.3 State-Level Expansion: Across 23 Districts


In terms of scale, the move from Madnapally to all 23 districts in the state represented a
quantum leap, as it involved a move from 49 to 1500 PHCs. However, the focus was not on
individual PHCs, but on the district capital, where data from all the PHCs in the district were
consolidated. In effect, it was necessary to scale up from 12 hubs to 23 district locations,
representing a vast geographical area. Because the allocated time for the project was very
limited (only 4 months), as was the budget, only three new trainers were hired, and each
trainer was allocated two districts, with responsibility to split time equally between them.
This phase involved a dramatic expansion in both geographical scope and technical
complexity. The limits of the Access database in the existing software became clearly
evident, and the whole database had to be ported to MySQL. In the district, the data were
still entered into the existing software; then an export file was created in text format in
each district, which was then sent by e-mail to the state capital, where the lead developer
imported the e-mailed file into the MySQL database with the use of a utility developed
for this purpose. A server space was rented in Singapore, and the state database was Web
enabled with DotNet, thus providing the functionality for users to view statewide reports.
Once the data for all the districts for 1 year were entered, the database size became 634 MB
(after compacting) and included more than 3.5 million records. The size of the database
made the system access very slow, a problem magnified by the relatively slow connectivity
that was available (32 Kbps).
Although the technical systems were relatively successfully scaled up in a short period
of 4 months, the scaling has been far more complex institutionally, for example, in getting
the state authorities to accept and use the systems. This complexity was further magnified
by the announcement of state elections in January. First the project was scaled down from 6
to 4 months, and second, the uncertainty of the elections made the officers feel ambiguous
about their own futures, contributing to their lukewarm interest toward the project. In May
2004, the election results were announced and the ruling party was voted out of power, with
the CFW being moved to another department. As a result of these changes, the systems
Information Technology for Development DOI: 10.1002/itdj
194 SAHAY AND WALSHAM

developed were without any political owner, although they were technically scaled up to
cater to the needs of the entire state. The systems were presented to the new Commissioner,
who required a fair amount of time to orient himself with the various initiatives ongoing in
his department before making any decision.
As an epilogue to this case story, the proposal for the state expansion was subsequently
accepted in July 2005, but in a revised form. An MoU was signed between HISP India and
the State Department to implement the Integrated Health Information System (IHIMS),
which was a combination of the FHIMS and DHIS. The MoU outlined that the IHIMS
would be implemented with the use of two models in the districts of Chittoor and Nalgonda,
respectively. While in Chittoor, the IHIMS would be installed in all 84 PHCs; in Nalgonda
the installation would be at the DMHO office. This project was subsequently evaluated
in December 2005, and HISP India was asked to submit another proposal to replicate the
Chittoor model in the rest of the state. This proposal has been submitted, and the State had
not yet made a decision at the time this article was written.

4. CASE ANALYSIS
The case analysis is presented in this section around two key questions: What is being
scaled? How is it being scaled?

4.1 What Is Being Scaled?


The three stages of the case represent three increasing levels of scale of multiple and
interconnected aspects. There is the increasing geographical scope and numbers of clinics
from 9 PHCs located in about 2,000 sq km (Kuppam) to 46 PHCs over 8,000 sq km
(Madnapally) to 1,500 PHCs over 15,000 sq km. The increasing geographical scope of the
operations was, paradoxically, accompanied with a reducing amount of time given allotted
for the project (1 year to 6 months to 4 months in the three phases, respectively). The three
maps given (Figures 2–4) depict the three respective levels of geographical scope: Kuppam,
Madnapally, and the whole state.
The increase in scope was accompanied with a magnification of the technical complex-
ity of the systems. From the stand-alone report-generation applications in Kuppam, there
was a requirement to integrate with FHIMS in Madnapally. In the state, the need was to
develop a Web-based application hosting the state database, which required a move from
Access to MySQL, and hosting the system on a Singapore-based server. The increasing
size of the database and the need for Internet access placed additional demands on network
connectivity and reliable power supply. To deal with the increasing scale, the implemen-
tation team had to be scaled up in terms of both numbers of people and skill levels. For
example, it was necessary to hire people with hardware maintenance skills to deal with the
hardware problems in Madnapally, and the state implementation required people with Web-
development and MySQL capabilities. All the members, in addition, needed to be oriented
to the HISP approach, which emphasizes a social-science perspective to implementation.
Being sensitive to people and organizational issues is not a quality that comes naturally to
computer-science graduates from Indian institutions, which typically emphasize technical
skills.
As the case progressed, there were also increasing political complexities. Whereas the
implementation in Kuppam could manage in a relatively stand-alone manner with support
Information Technology for Development DOI: 10.1002/itdj
SCALING OF HEALTH INFORMATION SYSTEMS IN INDIA 195

Figure 2 Distribution of nine primary health centers around five Mandals in the Kuppam sector,
Chittoor District.

Figure 3 Location of primary health centers in Madnapally Revenue Division, Chittoor District.
Madnapally district consists of three sectors—Madnapally, Piler, and Kuppam.

Information Technology for Development DOI: 10.1002/itdj


196 SAHAY AND WALSHAM

Figure 4 State map of Andhra Pradesh containing 23 districts, including Chittoor.

from the CM’s office, the Madnapally location had to deal with other divisions that had their
own functionaries. In the state, it was necessary to engage with all the state authorities, and
to ensure that the work was in line with the politics of the state (for example, the interest
of the state to promote the alternative FHIMS system), which also required alliances
with other agencies, for example, the technical developers of the FHIMS system. The
complexity was magnified by institutional factors such as the movement of functionaries
(for example, in Chittoor there were four different heads of the health department in
3 years), and political developments such as the announcement of elections, the results
of which led to the change of the ruling party and with it the transfer of the senior
health department functionaries. This required the renegotiation of alliances, an essentially
political process. This political process again had implications on the kind of skills required
of the implementation team members, as they frequently had to meet the bureaucrats and
politicians, make presentations, write proposals, and market the systems. As mentioned
earlier, these skills do not come naturally to technical staff, and also require maturity and
experience. A few core people in the HISP team were thus explicitly groomed by the
more senior staff to play these external roles. Although there was no predefined or set
methodology to do this, efforts were made by the senior staff to take these core people to
attend all the meetings, and also to help them contribute to the creation of presentations and
proposals.
So, the answer to the question of what is being scaled is not simple or a unidimensional
listing of factors. It can be described as a scaling up of complexity, best conceptualized
or represented as a heterogeneous network comprised of geography, numbers, technical
systems, data and databases, user capacities, trainers, and socio-technical practices such as
political negotiations that try to bring the network together.
Information Technology for Development DOI: 10.1002/itdj
SCALING OF HEALTH INFORMATION SYSTEMS IN INDIA 197

4.2 How Is It Being Scaled?


The present approach to scaling was multifaceted; some elements were planned, some were
unplanned, and others emerged from the changing political and institutional circumstances
that opened up new opportunities and closed off others. For example, the change of guard in
the Commissioner of Family Welfare opened up the opportunity for the health department
to reconsider the alternative as compared to FHIMS. However, the political changes also
resulted in the movement of a champion of the project from the Chief Minister’s office to
another one.
The present approach can be best described as one of cultivation, as described earlier,
which involved small incremental steps, sensitivity to the existing installed base, and
flexibility in dealing with the rapidly changing political and institutional conditions. For
example, the original plan was to start small, within a limited geographical setting, to
first try and understand the health system and to build the prototypes from the bottom up.
Starting in the nine PHCs in Kuppam was in line with this thinking. However, the next step
to expand to the Madnapally division was relatively unplanned and defined primarily by
the wishes of the CFW. There were also instances of the planned and unplanned working
together. For example, the requirement of placing the systems for the infant and maternal
mortality monitoring in all 23 districts was unplanned and arose from the wishes of the
CFW. However, given this opportunity, a conscious decision was made to place the DHIS
software in all the 23 districts simultaneously.
The installed base in this case is represented by the existing data items and report formats.
So, the first step was to adopt a participatory approach to see how the redundancies in the
data items could be reduced and an MDS developed. This process was in the interest of
the health workers, who perceived it would help to reduce their manual work; thus they
actively supported it. After this process, all the existing reports were automated, in exactly
the same format as the paper ones. This helped to allay fears that the reports would be
changed (which cannot be done easily in a hierarchical structure such as India), and again
was supported by the staff as it reduced the time spent by them every month in making
the reports manually on multiple paper forms. The definition of items in the MDS, as they
were standardized, also helped to take the systems relatively unproblematically to other
PHCs.
To deal with the challenge of increasing scale, in Madnapally a cluster approach was
adopted, thus enabling 12 computer sites to meet the needs of 46 PHCs. In the state, 23 sites
(of district head offices) were chosen to cater to the data needs of 1,500 PHCs. Although
some detail was lost in this process (for example, the SC-level data in the district database),
being able to provide statewide coverage appealed to the policy makers in the state. Placing
the system on the Web also helped to increase visibility and lend political legitimacy to the
state authorities, who wanted to have access to the latest technology (the Web in this case).
Similarly, the integration of the routine data to maps helped to provide visibility to a larger
geographical area, and also catered to the political needs of having the latest technologies.
Scaling problems arising from technical support were tackled by changing the kinds of
printers, and also by hiring a hardware engineer so as to reduce external dependence.
Needs for increased training were dealt with to some degree by preparing manuals that the
trainers could self-study to get up to speed with the basics of the software, and also use for
conducting training sessions for larger groups of people.
The cultivation approach can also be described in the process of developing the human
resources capacity of both the users and the implementation team. For example, the training

Information Technology for Development DOI: 10.1002/itdj


198 SAHAY AND WALSHAM

of users started with the very basics, for example, the different parts of the computer, how
to turn it on and off, and how to move the cursor. The trainers would then gradually
introduce the basics of software, including MS Office, for example, teaching the user how
to write a letter requesting leave using Word, or prepare a home budget with an Excel
spreadsheet. After this they were taught to use the DHIS software, starting from data
entry to report generation, and the more sophisticated users were taught how they could
conduct local analysis of the health data by using the software. Through this process of
incremental scaling up of training processes, the trainers also gradually built up their own
capacity.
Although a bottom-up and cultivation approach was used at first, this was soon comple-
mented or integrated with a top-down strategy of placing the systems only in districts and
the state capital. The bottom-up approach was necessary for the understanding of local-level
needs and for the development of relevant and robust applications. However, this in itself
would have been inadequate without the top-down approach, which helped the scale up
and thus provided a full coverage of the state, thus leading to a greater level of political
legitimacy and support. It is argued that an integration of both approaches is necessary for
successful scaling.

5. CONCLUSIONS: PRACTICAL AND THEORETICAL


CHALLENGES OF SCALING
Practically, scaling up of health information systems in PHCs is a very complex endeavor
for at least three reasons. One, there is the unique problem of “all or nothing.” This implies
that systems are practically useless for the health department unless the coverage of the
whole state is obtained. For example, to be able to compute the immunization coverage of
the state, data are required from all the reporting facilities collected in similar formats, and
using the same business logic for the calculation of the indicator. So, if work had continued
in Kuppam alone, the systems would have been of little use, as they would not have been
visible or relevant for the CFW. A second problematic condition for scaling concerns the
political nature arising from the confluence of interests of the donors, vendors, the political
leaders, and the bureaucrats. The PHC sector thus represents a battlefield comprised of
this multiplicity of interests and competition for large sums of money coming primarily
through donor agencies. This political nature cautions against simplistic scaling approaches
based on economic principles or numbers drawing upon traditional principles of network
externalities. A third condition concerns the continuous state of change and instability that
constitutes this sector. This instability arises from frequent political changes, revisions in
technological demands, and also changes in public health needs. For example, the escalation
of the current HIV/AIDS crisis is creating demands for new health initiatives that require
different kinds of data and reporting systems. Instability and change are thus inherent, and
they have direct implications on the scaling process.
Theoretically, unpacking the complexity of scale and scaling processes opens up a range
of challenging questions. There are a variety of dilemmas associated with scaling, relat-
ing, for example, to standardization and local customization, top-down and cultivation
approaches, treating people as whole humans versus as statistics, appropriate versus com-
plex technological solutions, and the all-or-nothing challenge particular to the PHC sector.
Although this article has alluded to some of these dilemmas, they need to be more carefully
examined and theorized upon, and insights must be developed through empirical analysis
in varied settings.
Information Technology for Development DOI: 10.1002/itdj
SCALING OF HEALTH INFORMATION SYSTEMS IN INDIA 199

Work to date reveals some of the implications for human resources capacity development.
A key implication concerns considering strategies for scaling of both the users of the systems
and also the members of the implementation team. With respect to the users, a cultivation
approach to training is useful, as it helps to enable more gradual transitions from their
existing ways of doing things to using the new technologies and associated approaches.
Another key implication to consider is that people involved in implementation should not
only be focused in technical terms, but should also have a sound understanding of the needs
for scaling up of systems, and how this process is not only about the software architecture
(which of course is crucial), but also about the escalation of complexity. This involves
considering institutional issues, politics, and the growing of the team itself. The team should
also have the ability to seize opportunities that may arise due to the occurrence of unplanned
events, and use it to their advantage to address the scaling challenges effectively. Another
point emphasized through this case, a point often neglected in many implementation reports,
is the manner in which the implementation team needs to be scaled up as the scope of
activities increase. This scaling is not only in terms of numbers, but also with respect to
the skills required. The associated challenges involve processes of communication and
coordination among the team members themselves so as to provide coherence to the overall
implementation effort.

REFERENCES
Appadurai, A. (1996). Modernity at large: Cultural dimensions of globalization. Minneapolis: Uni-
versity of Minnesota Press.
Braa, J., & Hedberg, C. (2002). The struggle for district based health information systems in South
Africa. The Information Society Journal, 18, 113–127.
Braa, J., Monteiro, E., & Sahay, S. (2004). Networks of action: Sustainable health information systems
across developing countries. MIS Quarterly, 28, 337–362.
Callon, M., & Law, J. (Eds.). (1986). Mapping the dynamics of science and technology. Sociology
of science in the real world. London: Macmillan Press.
Ciborra, C.U., Braa, K., Cordella, A., Dahlbom, B., Failla, A., Hanseth, O., et al. (2000). From control
to drift. Oxford: Oxford University Press.
Dahlbom, B., & Janlert, S. (1996). Computer future. [Mimeo]. Department of Informatics, University
of Gøteborg, Sweden.
Ellingsen, G., & Monteiro, E. (2003). Big is beautiful. Electronic patient records in Norway 1980–
2000. Methods of Information in Medicine, 42, 366–370.
Hanseth, O., & Aanestad, M. (2003). Bootstrapping networks, communities and infrastructures: On
the evolution of ICT solutions in healthcare. Methods of Information in Medicine, 42, 385–
391.
Hanseth, O., & Monteiro, E. (2004). Understanding information infrastructure. Retrieved June 1,
2004, from http://heim.ifi.uio.no/∼oleha/Publications/bok.pdf
Heeks, R., Mundy, D., & Salazar, A. (1999). Why health care information systems succeed or fail
(Working Paper Series Number 9). Manchester, United Kingdom: University of Manchester,
Institute for Development Policy and Management.
Information Infrastructure Technology and Applications (IITA). (1995). Interoperability,
scaling, and the digital library research agenda. Technical Report, Information In-
frastructure Technology and Applications Working group. Retrieved from http://www-
diglib.stanford.edu/diglib/pub/reports/iita-dlw/main.html
Kimaro, H.C., & Nhampossa, J.L. (2005). Analyzing the problem of unsustainable health information
systems in less-developed economies: Case studies from Tanzania and Mozambique. Information
Technology for Development, 11, 273–298.
Kohlemainen-Aitken, R. (2004). Decentralization’s impact on the health workforce: Perspectives of
managers. Human Resources for Health, 2, 1–11.

Information Technology for Development DOI: 10.1002/itdj


200 SAHAY AND WALSHAM

Latour, B. (1987). Science in action: How to follow scientists and engineers through society.
Cambridge, MA: Harvard University Press.
Latour, B. (1999). Pandora’s hope. Essays on the reality of science studies. Cambridge, MA: Harvard
University Press.
Monteiro, E. (1998). Scaling information infrastructure: The case of next generation IP in Internet.
The Information Society, 14, 229–245.
Robertson, R. (1992). Globalization: Social theory and global culture. London: Sage.
Rolland, K.H. (2003). Reinventing information infrastructures in situated practices of use. An in-
terpretive case study of information technology and work transformation in a global company.
Unpublished Ph.D. thesis, Department of Informatics, Faculty of Mathematics and Natural Sci-
ences, University of Oslo, Norway.
Rolland, K.H., & Monteiro, E. (2002). Balancing the local and the global in infrastructural information
systems. The Information Society, 18, 87–100.
Star, S.L., & Ruhleder, K. (1996). Steps toward an ecology of infrastructure: Design and access for
large information spaces. Information Systems Research, 7, 111–133.
Suchman, L.A. (1987). Plans and situated actions. New York: Cambridge University Press.
Walsham, G. (1993). Interpreting information systems in organizations. Chichester, United Kingdom:
Wiley.

Sundeep Sahay is a Professor at the Department of Informatics, University of Oslo, Norway. His
teaching and research interests are focused on issues of globalization and their relationship to work
practices, organizational arrangements, and the role of ICTs. A key current focus is on the design,
development and implementation of health information systems in the context of developing countries.
Geoff Walsham is a Professor of Management Studies at Judge Business School, University of
Cambridge, United Kingdom. His teaching and research is centered on the social and management
aspects of the design and use of ICTs in the context of both industrialized and developing countries. His
publications include Interpreting Information Systems in Organizations (Wiley, 1993), and Making a
World of Difference: IT in a Global Context (Wiley, 2001).

Information Technology for Development DOI: 10.1002/itdj

Das könnte Ihnen auch gefallen